Hello Just asking why do you want to split at paragraphs? Couldn't you just simply look for a word boundary and then split the columns? You could exclude tags from the split meaning wordboundaries cannot be inside <> that way no breaking of tags. Since the result is checked by a human they can easily change the appearance by including breaks as needed. So simply do your half split and than work your way back to the first word boundary. That should give a fairly balanced column.
Regards Stefan Langer